The purpose of this project is to conduct research in statistical methods with particular emphasis on those appropriate to the analysis of data from clinical and diagnostic trials and epidemiological studies of cancer. Many of the problems studied under this project arise from the consultative activities of the Section. In the past year this research has included an investigation of the Mantel-Haenszel procedure for early stopping in survival time studies, and investigation of tests for interaction for multiple 2x2 tables, determination of optimal treatment using initial covariate information in the analysis of clinical trail data, an investigation of situations in which unequal sample sizes are appropriate in comparative trials, development of a method for analyzing heterogeneity among indirect standardized mortality ratios, and computation of percentage points for the asymptotic distributions of the one and two sample Kolmogorov-Smirnov statistics for truncated and censored data. Preliminary work has been accomplished on comparison of various survival models incorporating covariate information and will be continued in the coming year.